|USFDC Home | USF Electronic Theses and Dissertations||| RSS|
This item is only available as the following downloads:
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001920134
007 cr mnu|||uuuuu
008 080104s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001839
Short, Eugene L.
Growth of oxide thin films on 4H- silicon carbide in an afterglow reactor
h [electronic resource] /
by Eugene L. Short.
[Tampa, Fla] :
b University of South Florida,
ABSTRACT: Oxide thin films were grown on 4H-SiC at low pressure and reduced temperatures using a remote plasma afterglow thermal oxidation method, achieving significantly faster growth rates than standard atmospheric furnace processes. The resulting SiO2/SiC structures were characterized by a non-contact corona-voltage metrology technique in order to extract capacitance-voltage information, to facilitate further analysis of the afterglow oxidation growth mechanism, and to determine the electrical behavior of defects. In addition, mass spectrometry experiments revealed the concentration of nitric oxide species in the afterglow reactor gas exhaust produced by the cracking of N2O molecules in the microwave plasma discharge. Oxidations were performed on n- and p-doped epitaxial 4H-SiC wafers at growth temperatures between 700¨C and 1100¨C. The afterglow oxidation process was determined to be primarily in the parabolic growth regime, and thus rate-limited by diffusion processes.^ Analysis of the parabolic growth rate temperature dependence revealed a break in activation energy between 0.46 eV and 1.51 eV at lower and higher temperature ranges, indicating a change in the dominating oxidation mechanism. In the proposed transport-limited mechanism, afterglow oxidation was suggested to be rate-limited by parallel diffusion of atomic oxygen radicals and excited singlet oxygen molecules to the SiO2/SiC interface. An alternative stress-relief mechanism suggested that viscous flow of SiO2 could relieve compressive stress in the oxide above 960¨C. In this case, growth would be stress-limited at low temperatures and diffusion-limited at higher temperatures. Regardless of the exact mechanism or temperature range, the data developed in this work suggest that afterglow oxidation rates of 4H-SiC are faster than atmospheric growth rates mainly because significant quantities of atomic and excited oxygen are generated in the microwave discharge independent of temperature.^ Using flatband voltages and accumulation capacitance values extracted from C-V measurements, worst-case charge densities associated with the oxide-semiconductor interfacial region were estimated. The charged defects were found to exist in the 1012/cm2 range regardless of growth temperature or oxide thickness. The charged defects were attributed to interface traps which capture majority carriers while the SiC is electrically stressed into accumulation during measurement. It was suggested that the traps failed to emit their charges within the time of measurement, even when the semiconductor was swept into depletion, and thus caused a shift in the observed flatband voltage. Mass spectrometry analysis showed that no thermal cracking of gas species occurs in the furnace at the detection level of the measurement, but rather significant quantities of nitric oxide are produced by the cracking of N2O molecules in the microwave plasma discharge independent of furnace temperature.
Thesis (M.S.E.E.)--University of South Florida, 2006.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 60 pages.
Adviser: Andrew Hoff, Ph.D.
x Electrical Engineering
t USF Electronic Theses and Dissertations.
Growth of Oxide Thin Films on 4HS ilicon Carbide in an Afterglow Reactor by Eugene L. Short, III A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Electrical Engineering Department of Electrical Engineering College of Engineering University of South Florida Major Professor: A ndrew M. Hoff, Ph.D. Stephen E. Saddow, Ph.D. Kenneth A. Buckle, Ph.D., P.E. Date of Approval: October 20, 2006 Keywords: oxidation, remote plasma, non-c ontact, corona-voltage, mass spectrometry Copyright 2006, Eugene L. Short, III
Acknowledgments First and foremost I am pleased to express my gratitude to my major professor, Dr. Andrew Hoff, for his invaluable support, in sight and guidance. It is truly an honor working with him. I am also indebted to Dr. Elena Oborina for all of her helpful assistance, especially with electrical characterization. The staff at Semiconductor Diagnostics, Inc. contributed useful colla boration and technical support for the FAaST measurement tool. Robert Tufts and Richard Everly at the Nanomaterials and Nanotechnology Research Center provided assist ance with technical i ssues regarding the afterglow reactor and cleanroom. I was deli ghted to have Dr. Stephen Saddow and Dr. Kenneth Buckle serve on my committee. I w ould also like to thank each of the SiC research group members for their teamwork and support.
i Table of Contents List of Tables ii List of Figures iii Abstract v Chapter 1. Introduction and Motivation 1 1.1. Silicon carbide material propert ies and device applications 1 1.2. Plasma-assisted low-pressure oxidation of silicon carbide 2 Chapter 2. Background 3 2.1. Deal-Grove linear-parabolic mode l for the oxidation of silicon 3 2.2. Atmospheric thermal oxidation of silicon carbide 4 2.3. SiO2/SiC charges and capacitance-voltage characteristics 7 2.4. Plasma-assisted oxidation 11 2.5. Mass spectrometry: constructi on, operation and analysis 13 Chapter 3. Experimental Setup and Procedure 17 3.1. Afterglow oxidation reactor system description 17 3.2. Oxidation experimental details 18 3.3. Mass spectrometer experiments 20 3.4. Non-contact corona-voltage metrology 22 Chapter 4. Results and Discussion 26 4.1. Capacitance-voltage characteristics 26 4.2. Oxide thickness and growth regime 28 4.3. Activation energy and afterglow oxide growth mechanism 31 4.4. Flatband voltage and charge estimation 38 4.5. Mass spectrometry results 43 Chapter 5. Summary and Conclusion 47 5.1. Growth rate, activation energy a nd afterglow oxidation mechanism 48 5.2. Interface defects and trapped charge model 49 5.3. Microwave excitation effects on afterglow gas composition 50 5.4. Suggestions for future work 51 References 52
ii List of Tables Table 4.1 Activation energies rate constants and Arrhenius linear fit quality for afterglow oxide parabolic growth rates on 4H-SiC. 33
iii List of Figures Figure 2.1 Oxidant fluxes in the Deal-Grove model. 3 Figure 2.2 Typical charged defects associated with a SiO2/Si structure. 7 Figure 2.3 Energy band diagram of a SiO2/p-SiC structure show ing the effect of deposited corona charge. 9 Figure 2.4 Example C-V characteristic of a SiO2/n-SiC structure obtained by noncontact corona-voltage metrology. 10 Figure 2.5 Typical effects of interface trap s (a) and oxide fixed or oxide trapped positive charge (b) on p-type C-V characteristics. 11 Figure 2.6 Schematic diagram of a quadrupole mass spectrometer. 14 Figure 3.1 Afterglow oxidation system. 18 Figure 3.2 Afterglow oxidation process schedule. 20 Figure 3.3 Corona-voltage me trology instrumentation. 23 Figure 4.1 C-V characteristics for ptype 4H-SiC oxidized at 1100C. 26 Figure 4.2 C-V characteristics for ntype 4H-SiC oxidized at 1100C. 27 Figure 4.3 Oxide equivalent thickn esses of films grown on 4H-SiC. 28 Figure 4.4 Thickness-time dependence for afterglow oxidation of SiC. 30 Figure 4.5 Squared thickness vs. time dependence for afterglow oxidation. 30 Figure 4.6 Arrhenius plot of parabolic rate curves for afterglow oxides. 32 Figure 4.7 Composite rates resulting from two processes with different activation energies occurring either in parallel or in series. 34 Figure 4.8 Suggested temperature dependence of diffusion rates for oxidant species in the afterglow process. 36
iv Figure 4.9 Suggested rate-temperature depe ndence of stress-lim ited and stress-free growth in the afterglo w oxidation process. 38 Figure 4.10 Maximum flatband voltages ex tracted from C-V measurements. 39 Figure 4.11 Capacitance measured for oxide layers on 4H-SiC. 40 Figure 4.12 Worst-case charge densitie s estimated for oxides on 4H-SiC. 41 Figure 4.13 Energy band diagram of interf ace states in n-SiC capturing majority carriers during accumulation stress (a) a nd causing early depletion (b). 42 Figure 4.14 QMS concentration levels of NO+ and N2O+ species resulting from nonexcited N2O source gas at various furnace temperatures. 44 Figure 4.15 QMS concentration levels of NO+ and N2O+ species resulting from excited N2O source gas at various furnace temperatures. 45
v Growth of Oxide Thin Films on 4HS ilicon Carbide in an Afterglow Reactor Eugene L. Short, III ABSTRACT Oxide thin films were grown on 4H-SiC at low pressure and reduced temperatures using a remote plasma afterglow thermal oxi dation method, achieving significantly faster growth rates than standard atmospheric furnace processes. The resulting SiO2/SiC structures were characterized by a non-contact corona-volta ge metrology technique in order to extract capacitance-voltage informa tion, to facilitate further analysis of the afterglow oxidation growth mechanism, and to determine the electrical behavior of defects. In addition, mass spectrometry experime nts revealed the concentration of nitric oxide species in the afterglow reactor gas exhaust produced by the cracking of N2O molecules in the microwave plasma discharg e. Oxidations were performed on nand pdoped epitaxial 4H-SiC wafers at growth temperatures between 700C and 1100C. The afterglow oxidation process was determined to be primarily in the parabolic growth regime, and thus rate-limited by diffusion pro cesses. Analysis of the parabolic growth rate temperature dependence revealed a br eak in activation ener gy between 0.46 eV and 1.51 eV at lower and higher temperature range s, indicating a change in the dominating oxidation mechanism. In the proposed transport-limited mechanism, afterglow oxidation was suggested to be rate-limited by parall el diffusion of atomic oxygen radicals and excited singlet oxygen mo lecules to the SiO2/SiC interface. An alternative stress-relief
vi mechanism suggested that viscous flow of SiO2 could relieve compressive stress in the oxide above 960C. In this case, growth woul d be stress-limited at low temperatures and diffusion-limited at higher temperatures. Regardless of the exact mechanism or temperature range, the data developed in this work suggest that af terglow oxidation rates of 4H-SiC are faster than atmospheric growth rates mainly because significant quantities of atomic and excited oxygen are generated in the microwave discharge independent of temperature. Using flatband voltages and accumulation capaci tance values extracted from C-V measurements, worst-case charge densi ties associated with the oxide-semiconductor interfacial region were estim ated. The charged defects were found to exist in the 1012/cm2 range regardless of growth temperature or oxide thickness. The charged defects were attributed to interface traps which capture ma jority carriers while th e SiC is electrically stressed into accumulation during measurement. It was suggested that the traps failed to emit their charges within the time of meas urement, even when the semiconductor was swept into depletion, and thus caused a sh ift in the observed flatband voltage. Mass spectrometry analysis showed that no thermal cracking of gas species occurs in the furnace at the detection level of the measurem ent, but rather significant quantities of nitric oxide are produced by the cracking of N2O molecules in the microwave plasma discharge independent of furnace temperature.
1 Chapter 1. Intr oduction and Motivation 1.1. Silicon carbide material pr operties and device applications Silicon carbide is a compound semiconducto r material attractive for electronic device applications requiring high power, hi gh voltage and high frequency operation in corrosive and high temperature environm ents. Among the most important polytypes investigated for power devices has been 4H-SiC which is composed of alternating Si and C atoms in a hexagonal lattice structure. The 4H-SiC polytype has a large energy bandgap of 3.26 eV compared to 1.12 eV for Si and an intrinsic carrier concentration roughly 19 orders of magnitude smaller than th at of Si. Silicon carbide is particularly appealing for metal-oxide-semiconductor device applications because it is one of the few compound semiconductors which can be th ermally oxidized, forming a native SiO2 layer due to the presence of Si in the crystal ma trix. However, most pr actical SiC devices to date have been junction type, as MOS structures of desired quality ha ve not been realized due to a number of problems. Bulk crystal qua lity is poor since substrates are produced by a sublimation process at very high temperat ures with high metal contamination levels. Furthermore, growth of high-quality defect -free epitaxial films critical for device applications has not been achieved. Epitaxial processes either ge nerate or propagate defects from the substrate, re sulting in a low quality startin g material for oxidation. Oxide films on SiC are chemically difficult to form and require growth temperatures between 1000C and 1300C in standard atmospheric fu rnace processes. A major factor that
2 complicates the formation of SiO2 films on SiC is the presence of C in the semiconductor material, which ideally should be removed from the system and not incorporated into the interface or oxide film. Typical SiO2/SiC structures exhibit a broad range of electrical defects. Much empirical work has been perfor med in an attempt to reduce the amount of defects, with limited success. The exact stru ctural and chemical nature of the SiO2/SiC interfacial region, the or igin of the charged electrical de fects and the mechanism of oxide formation on SiC are currently unknown. 1.2. Plasma-assisted low-pre ssure oxidation of silicon carbide Plasma-assisted low-pressure thermal oxi dation of silicon carbide by the remote plasma afterglow method is an advantageous alternative to standard atmospheric furnace processes. Plasma processes have attractive flexibility because gaseous species can gain energy or be dissociated by the excited plas ma discharge independent of the thermal environment. Hence, remote plasma oxida tion processes exhibi t little temperature dependence compared to atmospheric processes. The reduced temperatures and high growth rates achieved by afterglow oxidation of SiC are very appealing from a device and process technology perspective. The low pr essures and reduced temperatures used in the afterglow oxidation process imply drastically different reaction kinetics than those in the traditional atmospheric oxidation mode l. Atomic and excited oxidant species generated in a plasma discharge are suspected to play critical role s in oxidizing reactions at the SiO2/SiC interface. Remote plasma afterg low oxidation combined with non-contact metrology techniques for electrical characteriza tion of oxide films are valuable tools for further improving and understanding the oxidation of silicon carbide.
3 Chapter 2. Background 2.1. Deal-Grove linear-parabolic model for the oxidation of silicon A widely accepted general model for the thermal oxidation of silicon was developed by Deal and Grove . It assume s three series fluxes of oxidant molecules: gas transport to the oxide surface, diffusion tr ansport through the ex isting oxide layer and reaction with silicon atoms to form a new layer of oxide at the SiO2/Si interface. The three oxidant fluxes of the D-G model, il lustrated in Figure 2.1, are assumed to be identical, in a steady stat e condition past any initial transient behavior. Figure 2.1 Oxidant fluxes in the Deal-Grove model .
4 After some approximations, the differential equation describing the growth rate of the oxide film was derived as dxo/dt = (kC/N)/(1+k/h+kxo/D), where xo is the total oxide thickness, k is the surface reaction constant, C is the equilibrium concentration of oxidant molecules in the oxide layer, N is the numbe r of oxidant molecules incorporated into a unit volume of SiO2, h is the gas-phase transport coefficient and D is the effective diffusion coefficient of oxida nt molecules in the SiO2 film. Solving the differential equation for oxide thickness with the initial condition xo = xi at t = 0 yields the quadratic relation xo 2+Axo = B(t+ ), where B = 2DC/N, B/A = (C/N)/(1/k+1/h) and = (xi 2+Axi)/B. Two limiting cases exist for small or large oxidation times and thicknesses. In the surface reaction limited regime for short times and thin films, xo = (B/A)(t+ ), where B/A is the linear rate constant. In the transport limited regime for long times and thick films, xo 2 = Bt, where B is the parabolic rate consta nt. The Deal-Grove linear-parabolic model successfully predicts thermal oxide growth rates on silicon ove r a wide range of temperatures, times and thicknesses. However, for oxidation by O2 molecules in the thin initial growth regime, experimentally obs erved growth rates and thicknesses are consistently higher than predicted by the linear-parabolic model. 2.2. Atmospheric thermal oxidation of silicon carbide Recent experimental work has focused on applying atmospheric oxidation methods to thermally form SiO2 films on silicon carbide mate rial. Although silicon oxidation technology has been advanced and refined over the decade s, there remains room for much improvement in the growth of both the SiC crystal material and oxide layers with quality interfaces before acceptable field e ffect devices can be achieved on silicon
5 carbide. Growth temperatures between 1000 C and 1300C are required to oxidize SiC, due in part to the higher energy of the Si-C bond compared with the Si-Si bond . Thus, SiC oxidizes slower than Si. Furthermore, b ecause 4H-SiC has a sma ller lattice constant than Si, an abrupt SiO2/SiC interface is believed to be theoretically impossible. The interface supposedly consists of a transiti onal region between SiC and stoichiometric SiO2, containing silicon sub-oxides (SiOx x < 2), silicon oxy-carbides (SixOyCz) and other structural and carbon-related de fects [3-15]. The thickness of the SiO2/SiC interfacial region is on the or der of 50 , compared to an abrupt 5 oxide interface on Si. As a result, oxidation of SiC produces high er interface defect densities than those achieved on Si. The role of carbon must be accounted for in the oxidation kinetics. A linearparabolic model is generally applied to Si C atmospheric thermal oxidation, assuming O2 (dry) or H2O (wet) molecules as the oxidizing sp ecies which transport through the existing oxide and react with Si or C at th e interface. While oxidant reactions with Si atoms should lead to the formation of additi onal oxide, their reactions with C atoms are believed to form byproducts such as CO or CO2 which either become incorporated into the defect-filled interf acial region or oxide bulk, or out-d iffuse through the existing oxide film and exit the structure. While the bulk oxide is generally stoichiometric SiO2, the exact composition and structure of the tran sitional interface region is not known. Although many speculative opinio ns exist, the mechanism and kinetics of the SiC oxidation process are no t presently understood. Standard atmospheric oxidati on of SiC typically uses an ambient of either dry oxygen (O2), water vapor (H2O) or pyrogenic steam (O2+H2) at growth temperatures
6 between 1000C and 1300C, depending on the desired growth rate. Below 950C, no thermal growth occurs on SiC in dry or we t oxygen ambient under standard atmospheric conditions. For example, a 45 minute pyroge nic steam oxidation at 1100C produced 180 of oxide on 4H-SiC . Silicon carbide oxidation is generally followed by a re-oxidation annealing step in dry or wet oxygen at a temperature around 95 0C [3, 16-18]. The low temperature is chosen so that no further oxidation occurs at the interface, and no additional carboncontaining byproducts are generated as a re sult. During re-oxidation, oxidant molecules are suspected to further react with carbon in the interface or oxide and the resulting oxycarbide species undergo diffusion outward thro ugh the oxide, desorbing from the oxide surface to the gas phase. Re-oxidation anneals at 950C may also allow the oxide to relax and relieve stress at the interface. Although some improvements in interface and oxide quality have been achieved by re-oxidation, residual carbonrelated defects and silicon sub-oxides are suspected to plague the interfacial region. Various post-oxidation anneals [3, 19-27] have been studied in an attempt to reduce interface trap densities (Dit). The anneals are typically performed at non-oxidizing temperatures, similar to re -oxidation annealing, and have included a variety of ambients such as nitrous oxide (N2O), nitric oxide (NO), nitrogen (N2), ammonia (NH3), hydrogen (H2), and argon (Ar), with mixed results. The nitrogencontaining ambients, particularly nitric oxi de, seem to be effective at reducing or passifying defects in the interf acial region, yielding lower Dit values. Multiple compositional studies have established th at the nitridation anneals (excluding NH3) incorporate nitrogen in the interfaci al region only, not in the oxide bulk.
7 2.3. SiO2/SiC charges and capacitanc e-voltage characteristics The SiO2/SiC structure contains an abundance of charged defects in the interfacial region and oxide which degrade electrical pe rformance. Among these defects are the four types of charges generally associated with an oxide-silicon structur e and illustrated in Figure 2.2. Figure 2.2 Typical charged de fects associated with a SiO2/Si structure . Interface trapped charge (Qit) and oxide trapped charge (Qot) are attributed to stationary defects in the interfacial regi on or oxide bulk, resp ectively. Fixed oxide charges (Qf) are caused by defects in the oxide very near the interface, and are typically positive. Mobile ionic charge (Qm) is attributed to positive ions such as potassium or sodium which migrate through the oxide layer in response to electric fi elds applied to the structure. Specifically for the oxide-SiC st ructure, a non-abrupt interfacial region transitioning between SiC and SiO2 is suspected to contain many electrically active defects, including silicon sub-oxides, silic on oxy-carbides and othe r carbon-related or structural defects. Carbon re lated defects are also suspec ted in the oxide near the interfacial region.
8 A non-contact metrology technique implemen ting sequential corona ion deposition steps and contact pote ntial difference (CPD) measurements has been utilized to extract capacitance-voltage responses for the el ectrical characterization of the SiO2/SiC structure and its associated charged defects. Th e amount of deposited corona charge (Qc) on the oxide surface is monitored. The CPD probe measures a potential, VCPD = ms+Vox+VSB, where ms is the probe-semiconductor workfunction difference, Vox is the potential drop across the oxide, and VSB is the surface barrier due to any space charge in the semiconductor. The VCPD reading changes as each dose of charge is added to the measurement point on the oxide surface, VCPD = Qc/C. The total capacitance is extracted after each charge dose from the relation between the deposited charge and the corresponding voltage shift, C = Qc/ VCPD. The capacitance of the structure will vary depending on the amount and polarity of de posited corona charge, the presence of electrically active defects in the oxide or interface, and any space charge region in the semiconductor. An energy band diagram of a SiO2/p-SiC structure is displayed in Figure 2.3, illustrating the effect of deposited corona charge.
9 Figure 2.3 Energy band diagram of a SiO2/p-SiC structure show ing the effect of deposited corona charge . The total capacitance of the structure is the series combination of the dielectric capacitance and the capacitance due to any space charge region in the SiC. Interface traps can contribute an extra capacitance added to the space charge capacitance. For descriptive purposes, now consid er the total capacitance of an oxide film grown on n-type SiC, with electrons as majority carriers. For a large amount of positive corona ions deposited on the oxide surface, electrons will be accumulated and energy bands will bend downward in the SiC near the interface. Negl ecting interface traps, the only capacitance in accumulation will be that of the oxide layer, Cacc = Cox. As negative ions are then added to compensate the positive charge on the surface, the net deposited positive charge decreases and produces a reduction in the oxide field. As the field reduces, electrons are repelled from the interface into the SiC, which becomes depleted with energy bands bending upward. At the transi tion between accumulation a nd depletion of majority
10 carriers near the surface of the SiC is the flatband condition, where no bending occurs in the energy bands. At the flatband condition, the only measured voltage is the workfunction difference ( ms) between the CPD probe el ectrode and the SiC, VFB = ms, assuming no charged defects are present in the oxide or interface. In depletion, a space charge region exists in the SiC whose capac itance is in series with the dielectric capacitance, resulting in a lower total capac itance. As the deposited charge becomes increasingly negative, the SiC depletion regi on widens and the energy bands are bent further upward. The decreasing space charge capacitance associated with a widening depletion region causes the total capacitance to decrease until a minimum capacitance is reached in deep depletion. The general capacitance-voltage response of a SiO2/n-SiC structure obtained by the non-contact corona-v oltage method is displayed in Figure 2.4. Figure 2.4 Example C-V characteristic of a SiO2/n-SiC structure obtained by noncontact corona-voltage metrology.
11 When taking system defect charges into consideration, the gene ral effects on a C-V measurement are basically twofold. First, interface traps will cause the C-V curve to stretch out around the flatband as the stru cture is swept between accumulation and depletion. This occurs because some of the deposited charges contribute to the filling or emptying of interface states at various ener gy levels in the bandgap, instead of further accumulating or depleting the SiC. Second, any oxide trapped charge or fixed charge will translate the C-V curve along the voltage axis effectively shifting the flatband voltage (VFB) from its theoretical value. A certain amount of de posited charge is required to compensate for the oxide charge and achieve flatband condition. The voltage shift due to the oxide charge is related by the oxide capacitance, V = Q/Cox. The effects of interface traps and positive oxide charge are shown in Figure 2.5 for a p-type C-V characteristic. Figure 2.5 Typical effects of interface traps (a ) and oxide fixed or oxide trapped positive charge (b) on p-type C-V characteristics . 2.4. Plasma-assisted oxidation Plasma-assisted growth or annealing of oxide films at low pressures is an appealing alternative to standard atmospheric proce sses. The principle advantage of such an
12 approach is that a significan t portion of the en ergy input required to drive a chemical process can be gained from electrons in a plasma discharge, instead of from thermal energy at the ambient process temperature. Since the production of reactive precursors, intermediates, or the final products are le ss dependent on thermal energy input, plasmaassisted processes can be performed at re duced temperatures. Oxide growth at lower furnace temperatures is attractive for severa l reasons, including lower process cost and less re-distribution of prior doping. Plasma-assisted processes at temperatures as low as 400C have been employed to grow oxide films on Si using oxygen radicals as the oxidizing species [31-33]. Nitroge n radicals have been used fo r treating oxides on Si  and SiC [34-35]. Oxide films have previously been grown on SiC processed in flowing afterglow of a remote plasma containing oxyge n with or without nitrogen species [2, 3637]. High growth rates at reduced temperatures and low pressure were achieved. Atomic and excited oxygen were suggested as dominant species involved in the oxidation of SiC. An important study  found activation ener gies of 0.5 eV for atomic oxygen reacting with either polymeric or graphitic ca rbon. Excited singlet molecular oxygen (O2 *) also showed a 0.5 eV EA for reaction with polymer ic carbon. Interestingly, O2 reacted instantaneously at room temperature with graphitic carbon, yielding an immeasurable EA (essentially zero). Atom ic (O) and excited (O2 *) oxygen species present in an oxygen plasma afterglow are suspected to be respons ible for the removal of carbon during remote plasma afterglow oxidation of SiC by reac ting with inte rfacial C atoms or carboncontaining defects in the oxide a nd then out-diffusing as CO or CO2.
13 2.5. Mass spectrometry: cons truction, operation and analysis The quadrupole mass spectrometer (QMS) is a measurement tool capable of quantitatively determining the composition of a gas mixture and the partial pressures of its comprising gas species. The mass spectrometer is used at a wide range of pressures for a variety of gas analysis applications in the semiconductor industry. Although commonly employed as a residual gas analyzer or leak checker for monitoring background gas levels and contamination in high and ultra-high vacu um systems, the QMS is also well suited for process gas evaluation at higher pressures. A typical QMS operates by ionizing gas molecules in a vacuum environment, selec ting ions of a certain mass-to-charge ratio (m/e) and measuring their abunda nce, which is proportional to the partial pressure of the gas molecules. The three primary components of a QMS an alyzer are the ion source region, the quadrupole mass filter and the detector, all of which are aligned sequentially in a cylindrical radially symmetric assembly. Th e ion source consists of a hot, currentcarrying filament held at a negative potenti al with respect to, a nd contained within, a cylindrical mesh source electrode. An emissi on current of hot electrons flows between the filament and source electrode. Electron impa ct ionization of gas sp ecies in the source volume creates a distribution of primarily positive ions which are accelerated into the mass filter by a negatively biased extracti on electrode with a central opening. After exiting the ionization region, ions travel in an ellip tical spiral path down the center of a quadrupole mass filter comprising four parall el cylindrical conducting rods spaced radially around a central axis The quadrupole rods are driven with a specific set of DC and AC voltages such that an oscillating el ectromagnetic field is established in the
14 volume between the conductors. For a given field, only ions with m/e ratios in a particular range will have a stable path along the center of the rods, and all others will undergo increasing oscillations until they im pinge on a rod surface and are neutralized. Both the center and width of the selected m/e passband are determined by the applied voltages. The amplitudes are ramped such that the quadrupole scans through the m/e spectrum. As select ions exit the quadrupol e mass filter they are sensed by a detector consisting of a Faraday cup with an optiona l secondary electron mu ltiplier. The Faraday cup generates an electric current equal to the incoming ion current. The multiplier provides gain prior to the Faraday cup by em itting a stream of electrons proportional to incoming ions, but much higher in magnitude. The detected current is assumed to be proportional to the partial pressure of the orig inal gas species. A schematic of a typical QMS analyzer is shown in Figure 2.6. Figure 2.6 Schematic diagram of a quadrupole mass spectrometer . Several issues can limit the performance of a QMS. Outgassing from the hot filament and surrounding surfaces, thermal cracking and chemical reactions at the filament surface, and electron stimulated de sorption of ions from electron impact on
15 surfaces all contribute to the background interf erence levels. In addition, the ion mean free path requirement limits the total pressure inside the QMS assembly to 10-5 Torr. A partial pressure reduction (PPR) system, such as a low conductance orifice or metering valve placed between the process gas chamber and the QMS, must be used to analyze gases in a low vacuum environment at pressures higher than 10-5 Torr. Typically, the entire QMS assembly is differentially pumpe d to an operational level of at most 10-5 Torr. A concern with PPR systems is that the gas composition on either side of the orifice is not equivalent due to the flow propertie s of different gas spec ies in the mixture. Another serious problem is the reduction in part ial pressure of the ga ses sampled from the process chamber. For instance, if a 1 Torr pr ocess is monitored with a PPR QMS, and the internal pressure of the QMS is 10-6 Torr, then the partial pressure of any gas species in the process chamber will be reduced by a factor of 106 inside the spectrometer, perhaps yielding pressures lower than the background species levels within the spectrometer. Care must be taken when analyzing ma ss spectral data. Background levels present in the QMS, in addition to va rying flow properties of gas sp ecies through the PPR orifice and finite pumping capability of the QMS pumps contribute to gas composition changes in the QMS relative to that inside the proce ss chamber. Furthermore, a single gas species in the ionization tube will yield a characterist ic distribution of fragments, or cracking pattern, in the resulting spect ral data. For instance, water will produce a cracking pattern with descending concen tration levels of H2O+ (18 amu), HO+ (17 amu), H+ (1 amu), O+ (16 amu), H2 + (2 amu), H3O+ (19 amu) and so on. As part of the spectral analysis process, measured or assumed fragment patterns of su spected gas species are subtracted from the overall data pattern. For illustration, consid er using a QMS to analyze an environment
16 containing O2 and N2O process gases. The measured i on current corresponding to m/e 16 amu will have contributions from the O+ fragment of O2 (~10% [O2 +]), the O+ fragment of N2O (~5% [N2O+]), and the O+ fragment of background H2O (~2% [H2O+]). A general attempt has been made to appl y the oxidation mechanism knowledge and process technology that have been developed successfully for silicon to the oxidation of the wide-bandgap compound semiconductor silicon carbide, with less than satisfactory results. Numerous defects exist in SiO2/SiC structures, and the exact oxidation mechanism is not presently understood. Remote plasma afterglow oxidation at low pressure and reduced temperatures offers an attractive, flexible and effective method for growing oxide films on SiC. A corona-volta ge non-contact metrology technique can be used as a quick, non-destructive means for ex tracting capacitance-vol tage characteristics from experimentally grown oxide films. In a ddition, a mass spectrometer is a useful tool for analyzing the equilibrium molecular concen tration in a vacuum environment. For this work, oxide films were grown on silicon carbi de in an afterglow reactor at various temperatures. The SiO2/SiC structures were electrical ly characterized using non-contact corona-voltage measurements. A mass spectrom eter was used to analyze the effect of microwave excitation on the gas composition of the afterglow reactor.
17 Chapter 3. Experimental Setup and Procedure 3.1. Afterglow oxidation reactor system description The afterglow reactor system used for this work is capable of providing standard atmospheric pressure oxidation growth conditions. A resistive heating furnace surrounding the 6 in. diameter quartz tube gr owth zone can achieve temperatures up to 1200C. The available process gases and co rresponding maximum flow rates are oxygen (O2) at 10 slm, nitrogen (N2) at 10 slm, a 19 : 1 mixture of N2 : H2 also known as forming gas at 10 slm, nitrous oxide (N2O) at 1 slm and argon (Ar) at 1 slm. Furthermore, with the use of an upstream microwave cavity and vacuum pumps, the afterglow system can operate under low pressure, remote plasma afterglow assisted growth conditions. The microwave source provides a continuous wave 2.45 GHz signal which is carried along a rectangular waveguide and inje cted by a metal rod into the excitation cavity. The cavity surrounds a quartz tube containing the flowing gases on the inle t side of the furnace zone. The microwave excitation creates and ma intains a plasma discharge comprising unexcited and excited molecular and atomic gaseous species, ions, electrons and photons. Charged species (ions and electrons) are prev ented from exiting the excitation cavity by an RF choke formed by a grounded metalli c tube that extends above and below the cavity. A right angle bend in the quartz tube between the microwave cavity and furnace zone prevents photon ra diation, particularly of UV freque ncies, from entering the thermal zone of the system. Some residual amount of UV radiation, howev er, does travel down
18 the walls of the quartz tubing toward the grow th zone and is blocked by a glass fitting that joins the glass plasma tube and the fu rnace tube. Only the plasma afterglow species consisting of excited and unexcited neutral mo lecules and atoms enter the growth zone. The forward and reflected power of the microw ave excitation are controlled by the cavity insertion rod length and tuni ng stubs in the waveguide. A diagram of the afterglow oxidation reactor is shown in Figure 3.1. Figure 3.1 Afterglow oxidation system . 3.2. Oxidation experimental details Each of the oxidation experiments was performed on 8 miscut 0001 (Si face) oriented 4H-polytype silicon carbide 3-inch di ameter wafers. All SiC wafers had epitaxial layers, either nor p-type, on n-type substr ates. Typically, at leas t two SiC wafers were processed in a given process run, one of each epi doping type. The pre-furnace cleaning protocol included a standard RCA clean with 60C basic and acidic solutions of hydrogen peroxide (H2O2), as well as pre-, mid-, and post-cl ean dips in dilute hydrofluoric acid
19 (DHF) to achieve a clean SiC surface without oxide, organics, particles and metals. The first RCA clean solution (Standard Clean 1, SC1) was a 6:1:1 mixture by volume of 18 M /cm de-ionized water (DIW), ammonium hydroxide (NH4OH) and H2O2, and is known to remove organics, particles and some metals . The second RCA clean solution (Standard Clean 2, SC2) consisted of a 6:1:1 mixture by volume of DIW, hydrochloric acid (HCl) and H2O2, and is known to remove metals. The detailed wet cleaning process was as follows: DIW rinse, DHF dip, DIW rinse, SC1, DIW rinse, DHF dip, DIW rinse, SC2, DIW rinse, DHF di p, DIW rinse. Following wet cleaning, the wafers were dried in nitrogen and loaded in a quartz boat wi th 3/16-inch spacing between wafers. Several 3-inch silicon shielding wa fers, subjected to the same pre-furnace cleaning steps, were also loaded in front of and behind the SiC wafers that were oriented with their Si faces directed toward the pumping end of the afterglow tube. All oxidation experiments had identical process parameters with the exception of the temperature at which the oxidation step was performed, which ranged from 700C to 1100C. Following cleaning, drying and placement in the quartz boat, the wafers were loaded into the afterglo w reactor under flowing N2 at a furnace temperature of 600C. After pumping the chamber down to 1 Torr pressure, the temp erature was held at 600C for 20 minutes in N2 : H2 (19 : 1) afterglow ambient w ith microwave forward power of 1100 Watts and reflected power of 0-5 Watts Next, the reactor was ramped up from 600C to the oxidation temperatur e with flowing argon ambient at a pressure of several Torr. The oxidation step consisted of 20 minutes of O2 : N2O (13.3 : 1.0) afterglow with 1100 Watts microwave forward pow er at 1 Torr pressure. N2O was included to enhance the production of atomic O . Following th e oxidation step, the system was slowly
20 ramped down from the oxidation temperatur e to 600C in flowing argon ambient at a pressure of several Torr. Finally, the reactor was brought up to atmospheric pressure and the wafers were unloaded under flowing N2 at 600C. The wafers were allowed to cool to room temperature in a cleanroom ambient a nd then the backside (C face) oxide was stripped by HF vapor etchi ng in preparation for subseque nt metrology. The wafers were then rinsed in DIW and dried in nitrogen. The afterglow oxidation process schedule is depicted in Figure 3.2. Figure 3.2 Afterglow oxi dation process schedule. 3.3. Mass spectrometer experiments The quadrupole mass spectrometer (QMS) used to analyze the afterglow reactor exhaust was a Spectra Instruments MultiQuad setup. The QMS comprised an open source residual gas analyzer 6-inch head assembly with multiplier, turbomolecular and roughing pumps, partial pressure reduction (P PR) metering valve, and the Multi-Quad user interface control console. The electron beam in the QMS analyzer functioned at an electron energy of 70 eV. A turbomolecular pump, backed by a mechanical pump,
21 evacuated the QMS analyzer to a total pressure of 1-210-6 Torr during measurements or when the system was in idle mode. The PPR metering valve served as a small, controllable orifice to maintain the pressure difference between the QMS analyzer (10-6 Torr) and the afterglow system exhaust (1 Tor r). The Multi-Quad control console allows the user either to view a sw eep of all masses within a range (subset of 1-200 amu), or to specify up to 11 specific indi vidual masses for recording higher resolution scans. A series of experiments was performed usi ng the QMS to analyze the exhaust of the afterglow reactor under varying conditions of furnace temperature, gas mixture composition and microwave excitation. The gas compositions used were pure N2, a 19:1 mixture of N2:H2, pure O2, pure N2O, a 4:1 mixture of N2:O2, a 13.3:1.0 mixture of O2:N2O, a 13.3:1.0:2.2 mixture of O2:N2O:N2, and a 13.3:1.0:2.1:0.1 mixture of O2:N2O:N2:H2, where mixture ratios were calculated from the flow rates of the source gases. Furnace temperatures of 400C, 800C, and 1000C were used for the QMS experiments. At each combination of ga s mixture and temperature, the following procedure was executed. First, the microwav e cavity was excited and tuned to a forward power of 1050 Watts and reflecte d power of 0-5 Watts, with the reactor pr essure at 1 Torr. A few minutes were allowed to pass in order to ensure that the gas flows and plasma afterglow content had reached steady state. Then, a QMS repetitive sweep in the mass range 1-50 amu was viewed on the Mul ti-Quad display and the PPR metering valve was adjusted so that the highest peak was just below the maximum scale partial pressure for the QMS, approximately 910-7 Torr. The total pressure inside the QMS was 1-210-6 Torr. After the 11 most abundant masses were selected, a consecutive series of 5 scans over the 11 masses was performed in the hi ghest sensitivity mode The 5 consecutive
22 scans required roughly 2 minutes to comp lete. After scanning the afterglow gas composition, the microwave source was turned off and 5 minutes were allowed to elapse to ensure that the non-excited gas mixture had reached steady state. Neither the PPR metering valve on the QMS input, nor the pump isolation valve on the afterglow system exhaust line, nor the source gas fl ow rates were adjusted. As a result, the internal pressure of both the reactor chamber and the QMS dropped slightly after the microwave was turned off, due to the cracking of molecues that had occurred in the excited state. The procedure for viewing the 1-50 amu range sc an, selecting the 11 most abundant masses, and executing the 5 consecutive high sensitiv ity scans over the 11 masses was repeated for the gas mixture without microwave excita tion. Subsequent analysis was performed on the raw mass spectrometry experimental data The non-excited data for the pure source gases (N2, O2 and N2O) were taken as reference cracking patterns at each furnace temperature. The reference patterns were then used to calculate the compositions of the excited and non-excited gas mixtures at the respective temperatures. 3.4. Non-contact corona-voltage metrology A modified Film Analysis and Substrate Testing (FAaST) tool from Semiconductor Diagnostics, Inc. uses non-contact characteri zation techniques to perform a variety of semiconductor and film measurements [36, 42-50] Primarily, the tool operates using an ion source to deposit charge on the sample su rface from a corona discharge in air, and a Kelvin (CPD) probe to obtain potential read ings from the sample surface. The wafer sample is held on to a motorized metallic chuck by appling vacuum through the chuck to its backside and positioned under the corona gun and CPD probe, as shown in Figure 3.3.
23 Figure 3.3 Corona-voltage metrology instrumentation . After the corona gun deposits a dose of charge on the surface, monitored by a proprietary method, the chuck moves so that the same surface site can be immediately measured by the CPD probe. For the purposes of capacitance-voltage (C-V) characteristics of oxide films, the sequence of depositing a monitore d dose of charge and then extracting a VCPD reading is performed repetitively at a single site on the wafer. As part of a C-V sweep, the polarity, size and quantity of charge doses are specified. In addition, an initial charge dos e of possibly different polarity and larger size can be specified in order to establis h a charge condition at the meas urement site prior to starting the sweep. The total capacitan ce is extracted after each do se from the relation between the deposited charge and the co rresponding voltage shift, (C = Qc/ VCPD), and is plotted versus VCPD, yielding a capacitance-voltage curve. For example, to achieve a C-V sweep from accumulation to depletion for an oxide film on an n-type substrat e, an initial large dose of positive charge would be deposite d to produce majority carrier charge accumulation in the n-type semiconductor surf ace, and then the C-V measurement would proceed with a series of small doses of appl ied negative corona charge, gradually shifting the semiconductor surface charge from accumulati on to depletion of majority carriers.
24 The FAaST tool, as it is configured for SiC measurements, possesses the capability to perform non-contact C-V measurements with th e sample either in ambient darkness or under ultra-violet (UV) illuminatio n provided by a UV, 370 nm, diode. For each of the experimentally grown oxide films, several C-V curves were obtained at a single point 15 mm from the wafer center, away from the major flat. Sweeps were performed in both directions (from and toward accumulation for the particular epi doping type of the semiconductor) and under both dark and li ght conditions, in order to investigate the flatband shifting associated with charging traps while stressing into accumulation. The accumulation portion of th e illuminated C-V curve was used for calculating the equivalent oxide thickness (E OT). The EOT for an oxide layer is the equivalent thickness of stoichiometric SiO2 ( r = 3.9) which would yield a measured accumulation capacitance value, Cacc = Cox = oxo/tox = 3.9 o/EOT hence EOT = 3.9 o/Cacc. Thicknesses were calculated from the light measurements because the UV exposure generated carriers in the SiC, effec tively eliminating any depletion region in the semiconductor and leaving the SiO2 film as the sole contributor to the measured capacitance. For comparison, EOT calcula tions were also performed from the accumulation portions of the dark C-V curves. Oxide films were grown on 4H-SiC epitaxi al layers in a series of isochronal oxidation experiments performed in an afterglo w reactor using identical growth times at various temperatures ranging from 700C to 1100C. Several capacitance-voltage responses for each of the SiO2/SiC structures were extrac ted from non-contact coronavoltage measurements in both dark and light ambient. A temperature dependence analysis of film thicknesses and growth rates extracte d from the accumulation portions of the light
25 C-V curves fueled speculations regarding the nature of the afterglow oxidation mechanism and rate-limiting processes. Calc ulations involving fl atband voltages and oxide capacitances extracted from the C-V curves were used to estimate worst-case charge densities associated with the oxide -semiconductor interfacial region. Also, mass spectrometry scans were taken from the afterglow reactor exhaust for varying combinations of source gas mixture, fu rnace temperature and mi crowave excitation. Cracking pattern analysis was performed in order to reveal the effect of microwave excitation on the equilibrium molecular conc entration of gas species in the afterglow ambient.
26 Chapter 4. Results and Discussion 4.1. Capacitance-voltage characteristics Non-contact capacitance-voltage (C-V) curv es were obtained for oxide films grown on nand p-type 4H-SiC epi layers at temperatures ra nging from 700C to 1100C. Example plots selected from the resulting C-V curves are displayed in Figure 4.1 and Figure 4.2 for p-type and n-type, respec tively, at 1100C growth temperature. Figure 4.1 C-V characteristics for p-type 4H-SiC oxidized at 1100C.
27 Figure 4.2 C-V characteristics for n-type 4H-SiC oxidized at 1100C. Several general trends were observed from the collection of C-V curves. For both pand n-type, the accumulation capacitance values decreased with increasing temperature. Smaller oxide capacitances (Cox) were measured at higher temperatures primarily because capacitance is inversely pr oportional to thickness, Cox = oxo/tox, and thicker films were grown at higher temperatures. Second, the flatband voltages (VFB) generally increased with temperature, toward more negative or pos itive values for por n-type, respectively. Finally, the C-V curve stretch-out became more prominent at higher growth temperatures. However, these trends do not necessarily imply an increased amount of charged defects and carrier traps in the SiO2/SiC interfacial region or in the oxide, because Cox consistently decreased as temperature increased, and Q CV. This means larger voltage shifts and stre tch-out could be measured for thicker oxides with similar
28 amounts of charge in the structure. It is also worth noting that changes in VFB were observed for different C-V curv es at a given temperature. The most negative or positive VFB value was measured after strongly accumula ting the por n-type SiC, respectively. The shifting flatband is suspected to be caused by charge trapping in accumulation. 4.2. Oxide thickness and growth regime Oxide film thicknesses were calculated from the accumulation capacitance values measured with and without UV light exposur e for both pand n-type samples. The resulting EOTs are shown in Figure 4.3 w ith respect to growth temperature. Figure 4.3 Oxide equivalent thic knesses of films grown on 4H-SiC. Oxide thickness had an increasing super-linear dependence on temperature, regardless of the varying epi doping type a nd light conditions. Alt hough the light p-type and both light and dark n-type EOTs were roughly comparable within 10 angstroms (),
29 the dark p-type EOT was consistently 3040 thicker. Even though majority carriers were accumulated in the p-type SiC near the oxide interface, a depletion region is thought to exist at the (p-)epi / (n+)substrate junction. The capacitance of this space charge region was in series with the capacitance of the oxi de layer, thus lowering the total accumulation capacitance of the structure in the dark, and yielding a larger EOT value, EOT = 3.9 o/Cacc, where Cacc = (Cox -1+Csc -1)-1. An examination of film thickness depe ndence on time gives insight into the afterglow oxidation growth regime. Knowing whether the afterglow oxide growth is primarily linear or parabolic corresponding to reaction or transpor t limited processes will prove helpful for further analysis. Thickne ss values for 20-minute oxidations at 800C and 1100C performed for this work were co mpared with longer afterglow oxidations previously reported at the same temperat ures . All experiments used an O2 : N2O oxidizing ambient at 1 Torr in an afterglow reactor. The resulting thicknesses are shown in Figure 4.4, for 20 and 120 minute growth times at 800C, and 20 and 60 minute oxidations at 1100C, where the initial oxide thickness was assumed to be 5 at zero growth time. Clearly the oxide thickness does not increase linearly with time, but rather saturates in an assumably pa rabolic fashion. For further cl arification, the square of thickness was plotted against time, which should yield straight lines for a parabolic thickness-time dependence. Line ar fits to the squared EOT data are shown in Figure 4.5, confirming that the square of thickness does indeed increase linearly with time.
30 Figure 4.4 Thickness-time dependence for afterglow oxidation of SiC. Figure 4.5 Squared thickness vs. tim e dependence for afterglow oxidation.
31 The observed parabolic thic kness vs. time dependence imp lies that the afterglow oxide growth process is primarily diffu sion limited. This seems plausible when considering that the afterglow process occurs at 1 Torr total ambient pressure. The thermal oxide growth rate is predicted to be proportional to the concentration of oxidant species diffusing through the oxid e layer , which is in tu rn proportional to the partial pressure of oxidant in the ga s phase creating a concentrati on gradient across the oxide. Atomic (O) and excited (O2 *) oxygen species are suspected to play important roles in the afterglow oxidation [2, 36-38]. They are believed to exist in the afterglow ambient at concentrations at least two or ders of magnitude smaller than that of molecular oxygen (O2), which itself is bounded by the 1 Torr total pressure. It follows that there is very little concentration gradient across the oxide to drive the diffusion of oxidant species, and thus a very small flux of oxidants through the oxide layer. From this analysis the afterglow oxidation is assumed to be rate-lim ited primarily by diffusion processes, and will be considered as predominantly parabolic with a negligible linear regime at the start of growth. 4.3. Activation energy and afte rglow oxide growth mechanism An Arrhenius temperature dependence anal ysis was applied to the thickness vs. temperature data to gain insight into the oxidation growth mechanis m. The growth rate was assumed to be parabolic, and was estimated as the ratio of the square of thickness to oxidation time, Rox = EOT2/tox, in 2/min. The parabolic rate (Rox) was modeled as a simple Arrhenius expression with a single activation energy, Rox = Roexp(-EA/kT), where Ro is a rate constant in 2/min, EA is the activation energy in electron volts (eV), k is
32 Boltzmann's constant, 8.6210-5 eV/K, and T is growth temp erature in Kelvin (K). The activation energy is a measure of the growth rate's dependence on temperature, and is characteristic of the dominating or rate -limiting process in the oxidation growth mechanism. A semi-logarithmic Arrhenius plot of the natural log of th e parabolic rate vs. inverse temperature is depicted in Figure 4.6. Figure 4.6 Arrhenius plot of parabol ic rate curves for afterglow oxides. The activation energy was extracted from the slope of each Arrhenius curve, which is -EA/k. There was a discontinuity in the Arrhenius slope around 900C, but the curves were approximately linear in the temper ature ranges of 700C-900C and 900C-1100C. The activation energies and associated rate co nstants calculated from linear fits to the Arrhenius curves in the two temper ature ranges, as well as the R2 quality measures of the
33 linear fits, are displayed in Table 4.1 for all doping types and am bient light conditions. The EA values for both doping types under UV light and the dark n-type were comparable, averaging 0.46 0.1 eV and 1.51 0.03 eV in the lower and higher temperature ranges, respectively. Table 4.1 Activation energies, rate cons tants and Arrhenius linear fit quality for afterglow oxide parabolic growth rates on 4H-SiC. p-type SiC n-type SiC dark light dark light EA (eV) 0.3383 0.5663 0.4507 0.3635 Ro (2/min) 3.255104 2.022105 6.337104 2.400104 700-900C R2 fit 0.9211 0.9831 0.9961 0.9906 EA (eV) 1.306 1.506 1.490 1.539 Ro (2/min) 4.927108 2.317109 1.876109 2.924109 900-1100C R2 fit 0.9998 1.000 0.9964 0.9915 The change in EA around 900C indicates that the rate-limiting oxidation process is different at higher and lower temperatures. The afterglow oxidation growth mechanics are not kinetically simplistic, so the growth rate cannot be adequately described by a simple Arrhenius expression with a single act ivation energy for the entire temperature range under discussion. Several alternative ex planations could possi bly account for the observed growth rate temperature dependence. The rate-limiting afterglow oxidation gr owth mechanism could perhaps be a composite of two parallel independent pro cesses, each dominating at higher or lower temperatures. The overall growth rate would be the sum of the rates due to the individual parallel processes, so that at a given grow th temperature, the composite rate would be dominated by the faster of the two parallel processes. Assuming that the individual rate curves of the parallel processes intersect within the temperature range under consideration, the process with lower EA would determine the overall growth rate at
34 lower temperatures, and th e process with higher EA would determine the overall rate at higher temperatures. This woul d result in a composite Arrhenius rate curve, similar to that observed for the afterglow oxida tion of 4H-SiC, with a break in EA at the intersection of the rate curves of the individual processe s. Theoretical predictions of both a parallel and a series composite rate curve, together with individual rate curves of the two dominating processes, ar e depicted in Figure 4.7. Figure 4.7 Composite rates resulting from two processes with different activation energies occurring either in parallel or in series . Given that the afterglow oxidation is probabl y transport-limited and primarily in the parabolic growth regime, consider that tw o parallel diffusion processes could dominate the growth rate. It remains to identify which diffusion process is rate-limiting in each of the two temperature ranges. Excited singlet oxygen (O2 *), atomic oxygen radicals (O) and
35 molecular oxygen (O2) are three plausible candidates for species involved in oxidizing reactions at the interface. All three species may independently transport in parallel from the gas phase through the existing oxide laye r to the interfacial region. The possible reactions at the interface must also be considered in th e kinetics discussion, even though the rate limiting step is assumed to be a diffusion process. Atom ic oxygen radicals are suspected to react readily with both Si and C  in the interfacial region, facilitating both the formation of new Si-O bonds and the removal of unwanted C in the form of oxycarbide byproducts. Excited O2 is suspected to react readily with polymeric C and instantaneously with gr aphitic C . Although O2 could also react with Si to form new oxide, it is more likely that most of the available O2 is consumed by reactions with any graphitic or non-graphitic C in th e interfacial regi on. Non-excited O2 is suspected to react less favorably with Si and even more relu ctantly with C in the interfacial region, compared to the other two species. However, O2 cannot be dismissed completely, since its concentration is two or ders of magnitude higher. O2 and steam are capable of producing oxide films on SiC if the temperature is above 950C, but growth rates are slow. For comparison, consider that a 90 minute atmospheric pyrogenic steam process at 1100C furnace temperature produced only 300 of oxide on 4H-SiC , whereas a 20 mi nute afterglow oxidation produced 370 at the same temperature. Despite the ambient being steam, which oxidizes faster than pure O2 , and despite the pressure being 760 Torr instead of 1 Torr, the atmospheric oxidation process produced a growth rate r oughly 6 times slower than the afterglow oxidation rate. Thus, O2 alone is not capable of achievi ng the growth rates observed in the afterglow process.
36 Atomic O and excited O2 generated in the microwave discharge are present in the afterglow ambient at much smaller concentrations than O2. However, these species are believed to participate in the oxidizing inte rface reactions, especia lly the removal of C , and significantly enhance th e growth rate as a result. Th e availability of excited O2 or atomic O at the interface is suggested to be the rate-limiting fact or in the afterglow oxidation process. The diffusion rates of th ese two species through the oxide must be addressed. It is suspecte d that the diffusion of O2 through the oxide layer has more temperature dependence than atomic O diffu sion, considering the smaller size of the atomic O radical. Figure 4.8 shows the suggest ed temperature dependence of diffusion rates for oxidants in the afterglow ambient, as well as the resulting composite rate curve which could dominate the af terglow oxidation process. Figure 4.8 Suggested temper ature dependence of diffusion ra tes for oxidant species in the afterglow process.
37 According to the proposed transport-domina ted mechanism, the ra te-limiting step in the afterglow oxidation process is the diffusion of atomic O w ith slower growth rates and small EA at low temperatures, and the diffusion of excited O2 with faster rates and large EA at high temperatures. A change in EA for the afterglow growth rate occurs at the intersection of the rate curves of the two diffusion processes. Drawing from the observed afterglow rate data, it follows that 0.46 eV is the EA of atomic O diffusion in the oxide and 1.51 eV is the EA of excited O2 diffusion. Furthermore, the rate curves of the O and O2 diffusion processes are suggest ed to intersect around 900C. Alternatively, consider that the viscosity of oxide might be responsible for the observed growth rate temp erature dependence and EA break. Viscous flow of oxide beginning around 960C could relieve stress in th e oxide and alter the growth mechanics. Oxide viscosity has a large activation energy and hence an abrupt transition from nonflow to viscous flow around 960C . At temperatures below 960C, no viscous oxide flow occurs so large compressive stresses ex ist in the oxide due to the spacing mismatch of molar volume between the two materials. Compressive stress in the oxide could limit growth of additional oxide by preventing oxidant diffusion or interface reactions . This would produce slower growth compared to a stress-free film. Furthermore, stresslimited growth would exhibit little temper ature dependence below 960C because the amount of stress present in the oxide does not vary strongly with resp ect to temperature. Above 960C, viscous flow of oxide relieves stress in the oxide. This allows interface oxidizing reactions to occur uninhibited by stress and hence could result in faster growth rates. Diffusion of oxidant species through the oxide is suggested to be the rate-limiting step, with more of a dependence on temperat ure compared to the stressed film growth.
38 According to the proposed stress-relief mech anism, the afterglow oxidation process is stress-limited with slower growth rates and small EA below 960C, and diffusion-limited with faster rates and larger EA above 960C. The suggested rate-temperature dependence of stress-limited and stress-fr ee afterglow oxide growth is depicted in Figure 4.9. Figure 4.9 Suggested rate-temperature de pendence of stress-lim ited and stress-free growth in the afterglow oxidation process. Regardless of the temperature range or oxidation mechanism, afterglow growth rates are higher than atmospheric ra tes because the critical O and O2 species are generated in significant quantities by th e microwave discharge independent of temperature. 4.4. Flatband voltage and charge estimation Analysis was performed using VFB and Cox values extracted from C-V measurements to estimate the net charge due to traps and defects near the SiO2/SiC
39 interfacial region. The flatband vol tage is related to charge by a simplified model, as VFB = ms(Q/Cox) for nand p-type doping, respectively, where ms is the metalsemiconductor work function difference be tween the CPD probe and 4H-SiC, where s is a function of the Fermi level in the material Q is the net charge associated with the interfacial region, and Cox is the capacitance per unit area of the oxide film and is a function of the film thickness (tox) and stoichiometry ( r). The curves with largest VFB were measured following considerable st ressing into accumulation, and these were chosen in order to yield a worst-case es timate of charged defects. The maximum VFB values are shown in Figure 4.10 at various grow th temperatures for both pand n-type epi doping. Figure 4.10 Maximum flatband voltage s extracted from C-V measurements.
40 In general, the absolute value of VFB increased as the growth temperature increased, with a more pronounced effect in the p-type samples. Oxide capacitance values were extracted from the accumulation portions of the C-V curves measured under UV illumination. As seen in Figure 4.11, Cox decreased as temperature increased, mainly because oxide films were thicker. Figure 4.11 Capacitance measur ed for oxide layers on 4H-SiC. Values for the workfunction difference ( ms) between the gold CPD probe and 4HSiC were roughly estimated. Doping densities in the SiC were assumed to be around 1016 cm-3 corresponding to Fermi leve ls approximately 100-150 meV from the conduction and valence band edges for nand p-type epi-layers, respectively. The parameters used were the workfunction of gold, the workfunction of th e platinum reference us ed to calibrate the gold electrode, the electron affinity and bandga p of 4H-SiC, and the nand p-type Fermi
41 levels. The estimated ms values for nand p-type doping were 1.5 V and -1.0 V, respectively. The flatband equation was then solved for Q. The resulting worst-case charge densities are plotted in Figur e 4.12 with respect to temperature. Figure 4.12 Worst-case charge dens ities estimated for oxides on 4H-SiC. Some variation with temperature was obs erved, and charge densities for p-type were consistently higher than for n-ty pe. However, charges on the order of 1012/cm2 were generally present regardless of growth temperature or film thickness. Such a large areal density of charged defects is a significant obstacle to the successful oxidation of silicon carbide which needs to be overcome in the future. These charge traps are probably positioned in the interfacial region or fixed in the oxide layer near the interface, considering that if these de fects were distributed throughou t the oxide, th e density of charges would depend heavily on film thickness.
42 In view of the data, it seems plausible that the charged defects under discussion might be majority carrier traps in the interfacial region with energy states distributed in the near vicinities of the 4H-SiC conduc tion and valence band edges. These interface states apparently are generated without si gnificant dependence on oxidation temperature or film thickness. Assumed to be initially neutral, the interface traps are suggested to capture majority carriers during accumulation stressing and retain their charge when the semiconductor is no longer accumulated. This means that, once filled, the interface states do not emit their carriers even when returnin g past the Fermi level. The charges due to trapped majority carriers at the interface cause the semic onductor to become depleted earlier in the sweep from accumulation to depletion, with a corresponding shift of VFB. An illustrative energy band diagram is depict ed in Figure 4.13 for an n-type epi-layer. Figure 4.13 Energy band diagram of inte rface states in n-SiC capturing majority carriers during accumulation stress (a) and causing early depletion (b).
43 In the proposed model, interface states near the conduction band of the n-type semiconductor are initially positioned above the Fermi level and they are empty and neutral. Under an applied positive stress fiel d the energy bands are bent downward in the oxide and in the semiconductor as electrons ar e accumulated at the interface. As a result, the interface trap le vels near the conduction band fall below the Fermi level and they capture electrons and become negatively charged. The traps remain filled and retain their negative charge despite their energy levels rising above the Fermi level during the upward band-bending that occurs as the semiconductor is swept out of accumulation. The newly trapped negative charges at the in terface cause the semiconductor to become depleted prematurely, with a corresponding shift of VFB to a larger positive value following accumulation charging. 4.5. Mass spectrometry results The mass spectrometry experimental data was analyzed in order to reveal the effect of microwave excitation on the gas species co mposition of the afterglow reactor exhaust at various furnace temperatures. Only mol ecular species concentrations could be evaluated, identified by their mass-to-charge ratio (m/e). Due to the inherent limitations of mass spectrometry, excited and non-ex cited gaseous speci es could not be distinguished. Furthermore, atomic species c oncentrations could not be detected because of the higher pressures of atomic fragment s resulting from molecules cracking in the QMS. Thus, the excited and atomic species of oxygen and nitrogen e xpected to exist in the afterglow ambient were not characterized. Only equilibrium molecular densities in the afterglow reactor gas exha ust could be detected.
44 The most abundant molecular species in the analyzed mixtures were N2 (28 amu), NO (30 amu), O2 (32 amu) and N2O (44 amu). Particular emphasis was placed on investigating the concentr ation of nitric oxide (NO) molecu les, as NO is suspected to play an important role in reducing or passifying defects in the interfaci al region of oxidized SiC [3, 19-21, 27]. Since nitric oxide species are k nown to be produced from the cracking of N2O molecules [55-56], both NO and N2O species concentration levels in the QMS are of interest. Selected results are sh own in Figure 4.14 and Figure 4.15 for N2O source gas analyzed at various furnace temperatures without and with microwave excitation. Figure 4.14 QMS concen tration levels of NO+ and N2O+ species resulting from nonexcited N2O source gas at various furnace temperatures.
45 Figure 4.15 QMS concen tration levels of NO+ and N2O+species resulting from excited N2O source gas at various furnace temperatures. For non-excited N2O source gas, the NO+ concentration level is smaller than the N2O+ peak by a factor of 4. With micr owave excitation, however, the NO+ peak is larger than the N2O+ level by a factor of 8. Practically no variation of gas composition occurred over the furnace temperature range of 400 C-1000C. The absence of temperature dependence is a strong indication that negligible thermal cracking occurs as species travel through the furnace zone. For non-excited N2O source gas, this implies that the measured peak of NO+ species at 30 amu is due en tirely to the cracking of N2O molecules inside the ion source of the QMS assembly. In other words, the level of NO+ ions is 0.24 times that of N2O+ ions in the QMS fragment distribution of N2O, but the concentration of NO molecules in the reactor gas exhaust is too sm all for quantitative det ection. In the case of
46 N2O source gas with microwave excitation, the measured NO+ peak is 8 times larger than the N2O+ peak, and 32 times larger than expected from the NO+ ion fragment of N2O cracking in the QMS ion source. This is dir ect evidence for the production of NO species as N2O molecules are cracked in the microwave plasma discharge. Thus, nitric oxide species are not generated by thermal cracking of N2O molecules in the furnace, but rather are produced in significant quantities by N2O cracking in the microwave discharge independent of furnace temperature. Nitric oxide contributes to the interfacial quality of oxides grown on 4H-SiC by the afterglow method.
47 Chapter 5. Summary and Conclusion Oxide films were grown on 4H-SiC in a remote plasma afterglow reactor at low pressure and various temperatur es, achieving fast growth rates compared to atmospheric oxidation. Multiple capacitance-vo ltage responses of the SiO2/SiC structures were extracted from non-contact corona-voltage measurements and subsequently analyzed. The afterglow oxidation was determined to be primarily in the parabolic growth regime, indicating a diffusion-limited process. A break in activation energy was observed between the lower and higher temperature ranges, implying different rate-limiting processes. A transport-limited mechanism was proposed, in which the parallel diffusion processes of O and O2 dominate the growth rate. An alternative stress-relief mechanism addressing oxide viscosity was proposed, in wh ich low temperature growth is limited by compressive stress in the oxide, whereas grow th at high temperatures is limited by diffusion of oxidant species to the interface. Regardless of the exact mechanism or growth temperature, the fast afterglow oxidatio n rates were attributed to the significant quantities of O and O2 species generated in the micr owave discharge independent of furnace temperature. Worst-case charge densi ties of interfacial defects were estimated from maximum flatband voltages and oxide capacitances, yielding charges on the order of 1012/cm2 regardless of growth temperature or film thickness. The defects were proposed to be majority carrier traps in the interfacial region which become charged in accumulation and retain their charge despite returning past the Fermi level, producing a
48 flatband voltage shift. Mass spec trometry analysis revealed neglible thermal cracking of gas species in the furnace, but rather signifi cant quantities of nitric oxide were produced by N2O cracking in the microwave discharge independent of furnace temperature. 5.1. Growth rate, activation ener gy and afterglow oxidation mechanism Equivalent oxide thicknesses were extracted from the accumulation portions of C-V measurements. Film thickness increased with oxidation temperature for the isochronal growth experiments. By comparing thicknes ses with previously reported afterglow oxidations at longer growth times, the afterglow oxidation process was determined to occur primarily in the parabolic regime, implying a diffusion limited process. An Arrhenius temperature dependence analysis wa s applied to the parabolic growth rate, revealing a break in activati on energy around 900C, with EA values of 0.46 eV and 1.51 eV in the temperature ranges of 700 C-900C and 900C-1100C, respectively. The results suggest that a different rate-limiti ng process dominates the oxidation mechanism at higher and lower temperatures. Among the proposed explanations, competing parallel diffusion processes with different activation en ergies and intersecting rate curves could each dominate the oxide growth mechanism in the higher or lower temperature ranges, producing an EA break in the composite rate curve similar to that observed for the afterglow oxide growth rate on 4H-SiC. Oxidation due to O2 alone could not account for the high afterglow growth rate s. Highly reactive O and O2 species were suggested as crucial to the oxidizing interface reactions, espe cially the removal of C . The result is a faster oxidation rate which is lim ited by the availability of O or O2 at the interface. According to the proposed transport-lim ited mechanism, the diffusion of O and O2
49 through the oxide layer are the rate-limiting parallel processes that dominate the afterglow oxide growth. An alternative stress-relief mechanism suggested that the viscosity of oxide could account for the obser ved growth rate temperature dependence. An abrupt transition from nonflow to viscous flow of oxide around 960C could relieve physical stress in the oxide. Below 960C, oxidation could be limited due to oxide compressive stress, resulting in slow growth rates and little temper ature dependence. As viscous oxide flow relieves stress above 960 C, interface reactions proceed uninhibited by oxide stress. This yields faster growth ra tes with more temperature dependence, where oxidant diffusion to the interface is suggested to be the rate-limiting step. Regardless of the exact mechanism or temper ature range, the data sugge sts that afterglow oxidation rates of 4H-SiC are faster than atmospheri c growth primarily due to the significant quantities of O and O2 generated in the microwav e discharge independent of temperature. 5.2. Interface defects and trapped charge model A simplified model relating oxide capacitanc e and flatband voltage was used to estimate worst-case charge densities associ ated with the interfacial region. Oxide capacitances and maximum flatband volta ge magnitudes extracted from C-V measurements were decreasing and generall y increasing, respectively, with growth temperature. C-V curves with the largest fl atband magnitude were measured following electrical stress into accumulation. Charge density values were generally in the 1012/cm2 range, regardless of growth temperature or oxide thickness. The charged defects were attributed to majority carrier traps in the interfacial regi on. The interface states were
50 speculated to capture carriers while the semi conductor was stressed into accumulation, and fail to emit their charges within the tim e of measurement, even when their energy levels return past the Fermi level due to band-bending as the semiconductor surface is depleted. The trapped charges at the inte rface cause the semiconductor to deplete prematurely with an associated flatband shift to larger magnitudes following accumulation charging. 5.3. Microwave excitation effects on afterglow gas composition Mass spectrometry experiments were performed to analyze the effect of the microwave cavity excitation on the concentr ation of molecules exiting the afterglow oxidation reactor for various source gas mixt ures and furnace temperatures. The primary goal was to investigate the concentration of NO species produced by cracking of N2O molecules, since nitric oxide is believed to reduce or passify interfacial traps and carbonrelated defects for oxidized SiC. Negligib le temperature dependence was observed, indicating that thermal cracki ng of gas species in the furn ace does not occur. However, microwave excitation was shown to produce significant quantities of NO species by cracking of N2O molecules in the plasma discharge independent of furnace temperature. Although excited and atomic species were be yond the detection capability of the mass spectrometry analysis, the oxidation afterglow ambient can justifiably be expected to contain excited, non-excited, atomic and mol ecular varieties of oxygen and nitrogen in addition to nitrous and nitric oxide species. The O and O2 species, quantified in another study , are suspected to be critical for the oxidation pro cess and high growth rates, whereas nitric oxide is believed to enhance th e electrical quality of the interfacial region.
51 5.4. Suggestions for future work Isochronal experiments with c onstant growth times at va rious growth temperatures using oxidation mixtures other than O2 : N2O could produce changes in EA, growth rate and trapped charge density du e to variations in oxidizi ng ambient co mposition. This might shed further light on the afterglow oxidation mechanism and the origin of the charged defects. Oxidation experiments with varying gr owth times, particularly at higher temperatures, are required to investigate both charge and grow th rate dependence on time. Some of these were performed pr eviously at low temperatures . Experiments growing comparable oxide thicknesses at vari ous temperatures (requiring different growth times) would elimin ate film thickness as a cause of observed charge variation. Since the C-V characteristics are measured within minutes, an analysis of the time to empty for traps with energy levels > 150 meV from the band edges might be informative. The addition of optical emission spectroscopy (OES) actinometry measurement capability to the afterglow system coul d be used to quantify the atomic oxygen concentration in the oxidizing ambient [57-64]. Also nitric oxide titration experiments could reveal absolute concentrations of bot h atomic oxygen and nitrogen in the afterglow ambient .
52 References 1. B. E. Deal and A. S. Grove, "General Relationship for the Thermal Oxidation of Silicon", AIP Journal of Applied Physics, Vol. 36, No. 12, pp. 3770-3778, Dec. 1965. 2. A. M. Hoff, E. Oborina, S. E. Saddow and A. Savtchouk, "Thermal Oxidation of 4H-Silicon Carbide Using the Afterglow Method", Materials Science Forum, Vol. 457-460, pp. 1349-1352, 2004. 3. C. Raynaud, "Silica films on silicon carbide: a review of electri cal properties and device applications", Elsevier Journal of Non-Crystalline Solids, Vol. 280, pp. 1-31, 2001. 4. V. V. Afanasev, M. Bassler, G. Pens l and M. Schulz, "Intrinsic SiC/SiO2 Interface States", Phys. Stat. Sol. A, Vol. 162, pp. 321-337, 1997. 5. V. V. Afanasev, F. Ciobanu, S. Dimitr ijev, G. Pensl and A. Stesmans, "Band alignment and defect states at SiC/oxide interfaces", IOP Journal of Physics: Condensed Matter, Vol. 16, pp. S1839-S1856, 2004. 6. V. V. Afanasev and A. Stesmans, "Valence band offset and hole injection at the 4H, 6H-SiC/SiO2 interfaces", AIP Applied Physic s Letters, Vol. 77, No. 13, pp. 20242026, Sept. 2000. 7. R. Buczko, S. J. Pennycook and S. T. Pa ntelides, "Bonding Arrangements at the SiSiO2 and SiC-SiO2 Interfaces and a Possible Origin of their Contrasting Properties", APS Physical Review Letters, Vol. 84, No. 5, pp. 943-946, Jan. 2000. 8. S. T. Pantelides, G. Duscher, M. Di Ventra, R. Buczko, K. McDonald, M. B. Huang, R. A. Weller, I. Baumvol, F. C. Stedile, C. Radtke, S. J. Pennycook, G. Chung, C. C. Tin, J. R. Williams, J. Won and L. C. Feldman, "Atomic Scale Engineering of the SiC-SiO2 Interface", International Conference on Silicon Carbide and Related Materi als Proceedings, Oct. 1999. 9. E. Pippel, J. Woltersdorf, H. O. Olafss on and E. O. Sveinbjornsson, "Interfaces between 4H-SiC and SiO2: Microstructure, nanochemi stry, and near-interface traps", AIP Journal of Applied Physics, Vol. 97, 034302, 2005.
53 10. T. Ouisse and E. Bano, "Elect ronic Properties of the SiC-SiO2 Interface and Related Systems", IEEE, pp. 101-110, 1997. 11. G. V. Soares, C. Radtke, I. J. R. Baum vol and F. C. Stedile, "Morphological and compositional changes in the SiO2/SiC interface region induced by oxide thermal growth", AIP Applied Phys ics Letters, Vol. 88, 041901, 2006. 12. J. Dekker, K. Saarinen, H. Olafsson a nd E. O. Sveinbjornsson, "Observation of interface defects in thermally oxidized SiC using positron annihilation", AIP Applied Physics Letters, Vol. 82, No. 13, pp. 2020-2022, Mar. 2003. 13. R. N. Ghosh, S. Ezhilvalavan, B. Go lding, S. M. Mukhopadhyay, N. Mahadev, P. Joshi, M. K. Das and J. A. C ooper Jr., "Profiling of the SiO2-SiC Interface Using Xray Photoelectron Spectroscopy", Mate rials Research Society Symposium Proceedings, Vol. 640, pp. H3.7.1-H3.7.6, 2001. 14. L. I. Johansson and C. Virojanadara "Synchrotron radiat ion studies of the SiO2/SiC(0001) interfac e", IOP Journal of Physics: Condensed Matter, Vol. 16, pp. S3423-S3434, 2004. 15. T. Ohnuma, H. Tsuchida and T. Jikimoto, "Interface states in abrupt SiO2/4Hand 6H-SiC(0001) from first-prin ciples: Effects of Si da ngling bonds, C dangling bonds and C clusters". 16. M. Okamoto, M. Tanaka, T. Yatsuo and K. Fukuda, "Effect of the oxidation process on the electrical characteristics of 4H -SiC p-channel metal-oxide-semiconductor field-effect transistor s", AIP Applied Physics Letters, Vol. 89, 023502, 2006. 17. H. Yano, F. Katafuchi, T. Kimoto and H. Matsunami, "Effects of Wet Oxidation/Anneal on Interface Properties of Thermally Oxidized SiO2/SiC MOS System and MOSFET's", IEEE Transacti ons on Electron Devi ces, Vol. 46, No. 3, pp. 504-510, Mar. 1999. 18. M. Laube, T. Ohshima and G. Pensl, "Comparison of the Electrical Channel Properties between Dry and Wet Oxidized 6H-SiC MOSFET's Investigated by Hall Effect". 19. K. McDonald, R. A. Weller, S. T. Pantelides, L. C. Feldman, G. Y. Chung, C. C. Tin and J. R. Williams, "Characterization and modeling of the nitrogen passivation of interface traps in SiO2/4H-SiC", AIP Journal of App lied Physics, Vol. 93, No. 5, pp. 2719-2722, Mar. 2003.
54 20. G. Y. Chung, C. C. Tin, J. R. Williams K. McDonald, M. Di Ventra, S. T. Pantelides, L. C. Feldman and R. A. Weller, "Effect of nitric oxi de annealing on the interface trap densities near the band edge s in the 4H polytype of silicon carbide", AIP Applied Physics Letters, Vol. 76, No. 13, pp. 1713-1715, Mar. 2000. 21. C. Lu, J. A. Cooper Jr., T. Tsuji, G. C hung, J. R. Williams, K. McDonald and L. C. Feldman, "Effect of Process Variations and Ambient Temperature on Electron Mobility at the SiO2/4H-SiC Interface", IEEE Transa ctions on Electron Devices, Vol. 50, No. 7, pp. 1582-1588, Jul. 2003. 22. K. Fujihira, Y. Tarui, M. Imaizumi, K. Ohtsuka, T. Takami, T. Shiramizu, K. Kawase, J. Tanimura and T. Ozeki, "Cha racteristics of 4H-SiC MOS interface annealed in N2O", Elsevier Solid-State El ectronics, Vol. 49, pp. 896-901, 2005. 23. K. A. Ellis and R. A. Buhrman, "Nitr ous oxide (N2O) processing for silicon oxynitride gate dielectrics", IBM Journal of Research and Development, Vol. 43, No. 3, pp. 287-300, May 1999. 24. S. Dhar, S. Wang, J. R. W illiams, S. T. Pantelides and L. C. Feldman, "Interface Passivation for Silicon Dioxide Layers on Silicon Carbide", MRS Bulletin, Vol. 30, pp. 288-292, Apr. 2005. 25. G. Chung, C. C. Tin, J. R. Williams, K. Mc Donald, M. Di Ventra, R. K. Chanana, S. T. Pantelides, L. C. Feldman and R. A. Weller, "Effects of anneals in ammonia on the interface trap density near the band edges in 4H-silicon carbide metal-oxidesemiconductor capacitors", AIP Applied P hysics Letters, Vol. 77, No. 22, pp. 36013603, Nov. 2000. 26. Y. Hijikata, H. Yaguchi, S. Yoshida, Y. Ishida and M. Yoshikawa, "Effect of Ar post-oxidation annealing on oxide-4H-SiC interfaces studied by capacitance to voltage measurements and photoemission spectroscopy", AVS Journal of Vacuum Science and Technology A, Vol. 23, No. 2, pp. 298-303, Mar. 2005. 27. W. Wang, T. P. Chow, R. J. Gutmann, T. I ssacs-Smith, J. Williams, K. A. Jones, A. Lelis, W. Tipton, S. Scozzie and A. Agarwal, "Interface Properties of 4H-SiC/SiO2 with MOS Capacitors and FETs annealed in O2, N2O, NO and CO2". 28. B. E. Deal, Journal of the Electrochemical Society, Vol. 127, 1980. 29. "SDI Gate Dielectric Me trology C-V EOT and SASS", SD I Confidential, Ver. 3, Jun. 2004. 30. J. M. Aitken and E. A. Irene, Treatise on Materials Science and Technology, Vol. 26, Academic Press, 1985.
55 31. A. M. Hoff and J. Ruzyllo, "Atomic oxygen and the thermal oxidation of silicon", AIP Applied Physics Letters, Vol. 52, No. 15, pp. 1264-1265, Apr. 1988. 32. K. Sekine, Y. Saito, M. Hirayama and T. Ohmi, "Highly Reliabl e Ultrathin Silicon Oxide Film Formation at Low Temperat ure by Oxygen Radical Generated in HighDensity Krypton Plasma", IEEE Transactions on Electron Devices, Vol. 48, No. 8, pp. 1550-1555, Aug. 2001. 33. M. Togo, K. Watanabe, M. Terai, T. Yamamoto, T. Fukai, T. Tatsumi and T. Mogami, "Improving the Quality of Sub-1.5nm-Thick Oxynitride Gate Dielectric for FETs With Narrow Channel and Shal low-Trench Isolation Using Radical Oxygen and Nitrogen", IEEE Transactions on Electron Devices, Vol. 49, No. 10, pp. 1736-1741, Oct. 2002. 34. H. Yano, Y. Furumoto, T. Ni wa, T. Hatayama, Y. Uraoka and T. Fuyuki, "Radical Nitridation of Ultra-thin SiO2/SiC Structure". 35. Y. Maeyama, H. Yano, T. Hatayama, Y. Uraoka, T. Fuyuki and T. Shirafuji, "Improvement of SiO2/ -SiC Interface Properties by N itrogen Radical Treatment". 36. A. M. Hoff, "Growth and Metrology of Silicon Oxides on Silicon Carbide", MRS SiC symposium proceedings, 2004. 37. A. Tibrewala, "Thermal Oxidation of Single Crystal Silicon Carbide in Flowing Afterglow", Thesis, Department of Electri cal Engineering, College of Engineering, University of South Florida, 2002. 38. J. M. Cook and B. W. Benson, "Appli cation of EPR Spectroscopy to Oxidative Removal of Organic Materials", Journal of the Electrochemical Society, Vol. 130, No. 12, pp.2459-2464, Dec. 1983. 39. W. Umrath, "Analysis of gas at low pressures using mass spectrometry", Fundamentals of Vacuum Technology, Le ybold Vacuum Products and Reference Book, 2002. 40. W. Kern and D. A. Puotinen, "Cleani ng solutions based on hydrogen peroxide for use in silicon semiconductor technology", RCA Rev. 31:187-206, 1970. 41. J. E. Spencer, R. A. Borel, K. E. Linx wiler and A. M. Hoff, "Microwave apparatus for generating plasma afterglows", US Patent 4673456, 1987. 42. A. M. Hoff and E. Oborina, "Fast NonContact Dielectric Ch aracterization for SiC MOS Processing", ICSCRM, p. 1035, 2005.
56 43. A. Savtchouk, E. Oborina, A. M. Hoff and J. Lagowski, "Non-contact Doping Profiling in Epitaxial SiC", Material s Science Forum, Vol. 457-460, pp. 1349-1352, 2004. 44. P. Edelman, A. Savtchouk, M. Wilson, J. D'Amico, J. N. Kochey, D. Marinskiy and J. Lagowski, "Non-contact C-V measurements of ultra thin dielectrics", European Physical Journal Applied Physics Proceedings, 2003. 45. P. Edelman, A. Savtchouk, M. Wilson, J. D'Amico, J. N. Kochey, D. Marinskiy and J. Lagowski, "Non-Contact C-V Technique for high-k Applications", International Conference on Characterization and Metr ology for ULSI Technology Proceedings, Mar. 2003. 46. M. Wilson, J. Lagowski, A. Savtchouk, L. Jastrzebski and J. D'Amico, "COCOS (Corona Oxide Characterization of Semic onductor) Metrology: Physical Principles and Applications", ASTM Conference on Gate Dielectric Oxide Integrity Proceedings, 1999. 47. M. Wilson, J. Lagowski, A. Savtchouk, L. Jastrzebski, J. D'Amico, D. K. DeBusk and A. Buczkowski, "New COCOS (C orona Oxide Char acterization of Semiconductor) Method for Monitoring the Re liability of Thin Gate Oxides", M1Diagnostic Techniques for Semiconductor Materials and Devices Symposium, 196th Joint International Meeting of the Electrochemical Society Proceedings, Oct. 1999. 48. A. M. Hoff and D. K. DeBusk, M1 -Diagnostic Techniques for Semiconductor Materials and Devices Symposium, 196th Joint International Meeting of the Electrochemical Society Proceedings, Oct. 1999. 49. A. M. Hoff, D. K. DeBusk and R. W. Schanzer, SPIE Proceedings, Paper 3884-17, 1999. 50. P. Edelman, A. M. Hoff, L. Jastrzebsk i and J. Lagowski, SPIE Proceedings, Vol. 2337, pp. 154-164, 1994. 51. H. Z. Massoud, J. D. Plummer and E. A. Irene, "Thermal Oxidation of Silicon in Dry Oxygen: Accurate Determination of the Kinetic Rate Constants", Journal of the Electrochemical Society, Vol. 132, No. 7, pp. 1745-1753, Jul. 1985. 52. Richard C. Jaeger, "Introduction to Micr oelectronic Fabrication", Second Edition, Modular Series on Solid State Devi ces, Volume V, Prentice Hall, 2002. 53. E. P. EerNisse, Journal of the Elect rochemical Society, Vol. 35, No. 8, 1979.
57 54. D. B. Kao, J. P. McVittie, W. D. Nix and K. C. Saraswat, "Two-Dimensional Thermal Oxidation of Silicon: II. Mode ling Stress Effects in Wet Oxides", IEEE Transactions on Electron Devices, Vol. ED-35, p. 25, 1988. 55. Thermal cracking of N2O used to nitride thin ox ides on silicon, early 1980s. 56. L. A. Lipkin, Cree, "Method of N2O annea ling an oxide layer on a silicon carbide layer", US Patent 6610366, 2003. 57. R. E. Walkup, K. L. Sae nger and G. S. Selwyn, "St udies of atomic oxygen in O2+CF4 rf discharges by two-photon lase r-induced fluorescence and optical emission spectroscopy", AIP Journal of Ch emical Physics, Vol. 84, No. 5, pp. 2668-2674, Mar. 1986. 58. D. Pagnon, J. Amorim, J. Nahorny, M. T ouzeau and M. Vialle "On the use of actinometry to measure the dissociation in O2 DC glow discharges: determination of the wall recombination probability", IOP J ournal of Physics D: Applied Physics, Vol. 28, pp. 1856-1868, 1995. 59. P. Macko, P. Veis and G. Cernogora, "Study of oxygen atom recombination on a Pyrex surface at different wall temperatures by means of time-resolved actinometry in a double pulse discharge technique", IOP Plasma Sources Science and Technology, Vol. 13, pp. 251-262, 2004. 60. G. N. Kurtynina and D. K. Otorbaev, "Mass-spectrometry and actinometry of the low-pressure arc discharge with an eva porating graphite cathode ", IOP Journal of Physics D: Applied Physics, Vol. 30, pp. 2223-2228, 1997. 61. H. M. Katsch, A. Tewes, E. Quandt, A. Goehlich, T. Kawetzki and H. F. Dobele, "Detection of atomic oxygen: Improvement of actinometry and comparison with laser spectroscopy", AIP Journal of A pplied Physics, Vol. 88, No. 11, pp. 62326238, Dec. 2000. 62. N. C. M. Fuller, M. V. Malyshev, V. M. Donnelly and I. P. Herman, "Characterization of transformer coupl ed oxygen plasmas by trace rare gases optical emission spectroscopy and Langmuir probe analysis", IOP Plasma Sources Science and Technology, Vol. 9, pp. 116-127, 2000. 63. V. M. Donnelly, M. V. Malyshev, M. Schabe l, A. Kornblit, W. Tai, I. P. Herman and N. C. M. Fuller, "Optical plasma emission spectroscopy of etching plasmas used in Si-based semiconductor processing", IOP Plasma Sources Science and Technology, Vol. 11, pp. A26-A30, 2002.
58 64. T. M. Barnes, J. Leaf, S. Hand, C. Fr y and C. A. Wolden, "A comparison of plasma-activated N2/O2 and N2O/O2 mixt ures for use in ZnO:N synthesis by chemical vapor deposition", AIP Journal of Applied Physics, Vol. 96, No. 12, pp. 7036-7044, Dec. 2004. 65. P. Vasina, V. Kudrle, A. Talsky, P. Botos, M. Mrazkova and M. Mesko, "Simultaneous measurement of N and O densities in plasma afterglow by means of NO titration", IOP Plasma Sources Science and Technology, Vol. 13, pp. 668-674, 2004. 66. J. E. Spencer, R. A. Borel and A. M. Ho ff, "High Rate Photoresist Stripping in an Oxygen Afterglow", Journal of the Elec trochemical Society, Vol. 133, No. 9, pp. 1922-1925, Sept. 1986. 67. A. M. Hoff, E. Oborina, S. Aravamudha n and A. Isti, K2-Cleaning Technology and Semiconductor Device Manufacturing Symposium, 204th Joint International Meeting of the Electrochemical Society Proceedings, 2003. 68. A. M. Hoff, Materials Research Soci ety Symposium Proceedings, Vol. 815, pp. 189-198, 2004. 69. A. M. Hoff, T. C. Esry and K. Nauka, Solid State Technology, Vol. 39, No. 7, pp. 139-152, 1996. 70. H. Z. Massoud, J. D. Plummer and E. A. Irene, "Thermal Oxidation of Silicon in Dry Oxygen: Growth-Rate Enhancement in the Thin Regime: I. Experimental Results", Journal of th e Electrochemical Society, Vol. 132, No. 11, pp. 2685-2693, Nov. 1985. 71. H. Z. Massoud, J. D. Plummer and E. A. Irene, "Thermal Oxidation of Silicon in Dry Oxygen: Growth-Rate Enhancement in the Thin Regime: II. Physical Mechanisms", Journal of the Electroch emical Society, Vol. 132, No. 11, pp. 26932700, Nov. 1985. 72. E. A. Irene, "Models for the Oxidation of Silicon", CRC Critical Reviews in Solid State and Materials Sciences, Vol. 14, No. 2, pp. 175-223, 1988. 73. E. A. Irene, "Silicon oxidation studies: A revised model for thermal oxidation", AIP Journal of Applied Physics, Vol. 54, No. 9, pp. 5416-5420 Sept. 1983. 74. J. A. Costello and R. E. Tressler, "Oxi dation Kinetics of Silicon Carbide Crystals and Ceramics: I. In Dry Oxygen", Journal of the American Cera mic Society, Vol. 69, No. 9, pp. 674-681, Sept. 1986.
59 75. K. Christiansen, S. Christiansen, M. Albrecht, H. P. Strunk and R. Heibig, "Anisotropic oxidation of silicon car bide", Elsevier Diamond and Related Materials, Vol. 6, pp. 1467-1471, 1997. 76. R. K. Chanana, K. McDonald, M. Di Vent ra, S. T. Pantelides, G. Y. Chung, C. C. Tin, J. R. Williams and R. A. Welle r, "Fowler-Nordheim hole tunneling in pSiC/SiO2 structures", AIP Applied Physic s Letters, Vol. 77, No. 16, pp. 2560-2562, Oct. 2000. 77. R. Kosugi, K. Fukuda and K. Arai, "The rmal oxidation of (0001) 4H-SiC at high temperatures in ozone-admixed oxygen gas ambient", AIP Applied Physics Letters, Vol. 83, No. 5, pp. 884-886, Aug. 2003. 78. M. K. Das, "Recent Advances in 4H-SiC MOS Device Technology". 79. S. Wang, M. Di Ventra, S. G. Kim and S. T. Pantelid es, "Atomic-Scale Dynamics of the Formation and Dissolutio n of Carbon Clusters in SiO2", APS Physical Review Letters, Vol. 86, No. 26, pp. 5946-5949, Jun. 2001. 80. N. G. Wright, C. M. Johnson and A. G. O'Neill, "Mechanistic model for oxidation of SiC", Elsevier Materials Science and Engineering B, Vol. 61-62, pp. 468-471, 1999. 81. H. J. von Bardeleben, J. L. Cantin, M. Mynbaeva, S. E. Saddow, Y. Shishkin, R. P. Devaty and W. J. Choyke, "EPR Studies of SiC/SiO2 Interfaces in n-type 4Hand 6HOxidized Porous SiC". 82. H. J. von Bardeleben, J. L. Cantin, Y. Shishkin, R. P. Devaty and W. J. Choyke, "Microscopic Structure and Electr ical Activity of 4H-SiC/SiO2 Interface Defects: an EPR study of oxidized porous SiC". 83. J. Choi, S. Kim, J. Kim, H. Ka ng, H. Jeon and C. Bae, "Effects of N2 remote plasma nitridation on the structural and electrical character istics of the HfO2 gate dielectrics grown using remote plasma atomic layer deposition methods", AVS Journal of Vacuum Science and Technology A, Vol. 24, No. 4, pp. 900-907, Jul. 2006. 84. P. Tristant, G. Gousset, C. Regnier a nd J. Desmaison, "Modelling of a microwave flowing oxygen discharge: application to re mote plasma enhanced CVD of silica films", IOP Plasma Sources Scie nce and Technology, Vol. 11, pp. 241-247, 2002. 85. M. J. Pinheiro, G. Gousset, A. Granie r and C. M. Ferreira, "Modelling of lowpressure surface wave discharges in flow ing oxygen: I. Electrical properties and species concentrations", IOP Plasma Sources Scienc e and Technology, Vol. 7, pp. 524-536, 1998.
60 86. M. L. Passow, M. L. Brake, P. Lopez, W. B. McColl and T. E. Repetti, "Microwave Resonant-Cavity-Produced Air Discharges", IEEE Transactions on Plasma Science, Vol. 19, No. 2, pp. 219-228, Apr. 1991. 87. R. Etemadi, C. Godet and J. Pe rrin, "Phenomenology of a dual-mode microwave/RF discharge used for the depostion of silicon oxide thin layers", IOP Plasma Sources Science and Technology, Vol. 6, pp. 323-333, 1997. 88. M. Baeva, X. Luo, B. Pfelzer, T. Re psilber and J. Uhlenbusch, "Experimental investigation and modelling of a low-pr essure pulsed microw ave discharge in oxygen", IOP Plasma Sources Science and Technology, Vol. 9, pp. 128-145, 2000. 89. S. M. Sze, "Physics of Semiconductor Devices", Second Edition, John Wiley & Sons, Inc., 1981. 90. D. K. Schroder, "Semiconductor Material and Device Characterization", Second Edition, John Wiley & Sons, Inc., 1998. 91. E. H. Nicollian and J. R. Brews, "MOS (Metal Oxide Semiconductor) Physics and Technology", Bell Telephone Laboratories, Inc., 1982. 92. S. A. Campbell, "The Science and Engi neering of Microelect ronic Fabrication", Second Edition, Oxford University Press, 2001. 93. Robert F. Pierret, "Field Effect Devices", Second Edit ion, Modular Series on Solid State Devices, Volume IV, Addison-Wesley Publishing Company, 1990. 94. S. Wolf and R. N. Tauber, "Pro cess Technology", Second Edition, Silicon Processing for the VLSI Era, Volume 1, Lattice Press, 2000. 95. S. Mahajan and K. S. Sree Harsha, "Pri nciples of Growth and Processing of Semiconductors", The McGraw-Hill Companies, Inc., 1999. 96. J. D. Plummer, M. D. Deal and P. B. Griffin, "Silicon VLSI Technology, Fundamentals, Practice and Modeling", Pren tice Hall Electronics and VLSI Series, Prentice Hall, Inc., 2000.